We examine metrics for measuring clutter effectiveness on model-based automatic target recognition (ATR) systems with forwardlooking infrared (FLIR) sensors. The measure for clutter effectiveness proposed is the difference of two Kullback-Leibler distances between the idealized approximate probabilistic models without clutter and the real model containing clutter. We establish that occluding objects and clutter, when manipulated, do not present a fundamental challenge to model- based ATR systems if the model manipulated is an accurate representation of the obscuring clutter. However, if the obscurer is not manipulated, performance degrades in cases where the obscurer is an ''effective clutterer. To quantify the effect of clutter in ATR, estimation and detection problems are considered for rigid ground-based targets. For estimating the orientation of a vehicle, the Hilbert-Schmidt distance is employed.

To locate and recognize ground-based targets in forwardlooking IR (FLIR) images, 3-D faceted models with associated pose parameters are formulated to accommodate the variability found in FLIR imagery. Taking a Bayesian approach, scenes are simulated from the emissive characteristics of the CAD models and compared with the collected data by a likelihood function based on sensor statistics. This likelihood is combined with a prior distribution defined over the set of possible scenes to form a posterior distribution. To accommodate scenes with variable numbers of targets, the posterior distribution is defined over parameter vectors of varying dimension. An inference algorithm based on Metropolis-Hastings jump-diffusion processes empirically samples from the posterior distribution, generating configurations of templates and transformations that match the collected sensor data with high probability. The jumps accommodate the addition and deletion of targets and the estimation of target identities; diffusions refine the hypotheses by drifting along the gradient of the posterior distribution with respect to the orientation and position parameters. Previous results on jumps strategies analogous to the Metropolis acceptance/rejection algorithm, with proposals drawn from the prior and accepted based on the likelihood, are extended to encompass general Metropolis-Hastings proposal densities. In particular, the algorithm proposes moves by drawing from the posterior distribution over computationally tractible subsets of the parameter space. The algorithm is illustrated by an implementation on a Silicon Graphics Onyx/Reality Engine.

Many methods have previously been devised to estimate the relative amounts of chemicals present in a measured Raman spectrum. However, relatively little work has been done on developing physics-based probabilistic models for the measurement system. Drawing from previous work in astronomical image restoration, we model the acquired data based on the physics of two key components in our Raman instrumentation: the spectrometer and the charge-coupled device detector. Under this model, we derive Cramér–Rao lower bounds for the mixing coefficients of the target spectra. This bound is compared against the performance of several classification algorithms. The non-negative iteratively reweighted least-squares algorithm is seen to give performance that is nearly identical to the more computationally demanding expectation-maximization approach; this is true even at weak signal levels.

Most research regarding passive radar exploiting ‘illuminators of opportunity’, such as FM radio, has focused on detecting and tracking targets. This study explores adding automatic target recognition (ATR) capabilities to such systems. The ATR algorithms described here use the radar cross‐section (RCS) of potential targets, collected over a short period of time. The received signal model accounts for aircraft position and orientation, propagation losses, and antenna gain patterns. One proposed algorithm uses a coordinated flight model to estimate aircraft orientations, while a more sophisticated algorithm uses an extended Kalman filter to estimate the target orientations along with measures of uncertainty in those estimates. In both cases, the orientations are estimated using velocity measurements obtained from a tracking algorithm. The radar return of each aircraft in the target library is simulated as though each is executing the same manoeuvre as the target detected by the system. To improve the robustness of the result, the more sophisticated algorithm jointly optimises over feasible orientation profiles and target types via dynamic programming.